FastNeura Presents AI Model Generating ECG from PPG at JSAI2026

株式会社FastNeura announced research results at the 2026 Annual Conference of the Japanese Society for Artificial Intelligence (JSAI2026), held from June 8 to June 12, 2026, regarding an AI model that generates ECG from PPG.

In this research, they proposed a two-stage structure that introduces RR series as an intermediate representation, unlike conventional models that focus solely on waveform reproduction. They constructed a model that generates ECG while maintaining physiological consistency such as heartbeat timing and RR intervals.

Using the PPG-DaLiA dataset, they confirmed that the proposed method reduces heartbeat timing errors. The company aims to apply this as a higher-order state estimation technology using data obtainable from wearable devices.